基于支持向量机的模糊稳健优化设计方法

Fuzzy-robust Design Optimization Using Support Vector Machine

在模糊稳健设计中,需要采用随机模拟方法计算模糊概率和非线性约束函数,但计算效率很低。为此,提出了一种基于支持向量机的模糊稳健设计方法。采用支持向量回归机对模糊概率进行仿真计算,采用支持向量回归机或分类机作为非线性约束函数的替代模型,显著降低了模糊稳健优化设计的机时消耗。给出了新方法的具体算法步骤,并通过模糊稳健优化设计实例对所提出的方法进行了验证。

The stochastic simulation method is commonly used in fuzzy-robust design optimization to compute the values of fuzzy probability and nonlinear constraint functions,but its low computational efficiency is a problem to be solved.A support vector machine-based method was proposed for fuzzy-robust design optimization to solve this problem.Support vector regression（SVR） was used as a surrogate model of the stochastic simulation method for computing the fuzzy probability.Each of the nonlinear constraints was transformed into a problem of two-class classification,and then support vector classification（SVC） was used as its surrogate model to judge the feasibility of this constraint.By using these surrogate models,the computational efficiency of fuzzy-robust design optimization was distinctly improved.The arithmetic steps of the proposed method were given.An example was given to show that the proposed method is effective.